We're impatient, not lazy
One common argument against NLP is that most searchers aren't willing or able to type more than two words. This grossly underestimates our capabilities and attention spans, and it belies the experience of companies in a range of industries. Fixating on keyword versus natural language distinctions and query length, without first taking into account the context of the search behavior and searcher's objective, is near meaningless.
When the objective is to find a restaurant or a website whose URL you can't remember, shorter search queries often suffice. When the objective is more specific in nature and the content being searched has a higher frequency and density of likely keywords (think of your results if you type in "phone doesn't work"), the searcher is often presented with a long list of results to comb through for their actionable information. If offered an alternative that delivers searchers to information faster, most users will take it. If they are looking for a specific item or answer, they will gladly enter more explicit queries if it limits the number of post-search clicks required to get to the actionable information. Successful examples abound, from software companies that report customers are using NLP search to resolve complex support problems, to call center agents working with impatient customers who appreciate very targeted, excerpted results that completely eliminate the need for even a single clickthrough.
It is important to note that keyword search and NLP search are not mutually exclusive. Most NLP engines are equally effective at keyword matching. It's simply the case that the more specific the query, the more precise the answer. In action-oriented business environments, this is often the preferred usage scenario.
A new study from market research firm ServiceXRG confirms this behavior. Shoppers had a variety of options for finding the information they need to reach a purchase decision, yet 74.5 percent indicated that they start with the websites of the companies that produce the relevant products or services. According to study author Tom Sweeney, "The data suggest that these companies have the first opportunity to provide the tools and content to influence the shopping experience. Still, while company websites are the preferred destination for consumers embarking on a search, many come up short in meeting customer expectations: Only 44 percent of customers indicate that the information found on company websites meets their shopping and product research needs."
Where keyword searches fall short
There's a growing trend toward web self-help. There are no hold times or robotic agents -- there's 24/7 availability, and it allows people to answer questions they would not otherwise ask. The content available through support portals and in call centers is not nearly as disparate as you might find on the web. The density and frequency of product, company and industry keywords is much greater, so keyword-matching engines would return very long search result lists. What the user needs is an engine that can take context into account, and NLP is one of the best ways to capture the context of the search session and the intent of the searcher.
Look around and you will find numerous examples -- banking, insurance, telecommunications, retail electronics, etc. -- where keyword search has been replaced by NLP engines with more advanced features. It's not too hard to find examples to compare and contrast the two search experiences, even on .com sites. Compare T-Mobile against AT&T Wireless, for example.
Start with a specific need in mind (like determining which cell phone and calling plan to purchase), then compare the search experiences on each site. Start with basic keyword searches and compare the experience against how well you find the information you need to make a purchase decision. Then compare the experiences with longer, more explicit queries that align closer to what you are actually thinking (i.e. what is coverage in N.Y., phones under $100, phones with a camera, rules for rollover minutes, online bill payment options, cheap phones). The differences will be clear: parity on shorter searches and superior experience through NLP with longer queries.
So don't get caught in the hype. Don't be bamboozled by Google! Don't confuse internet search for an enterprise or business solution. Are your customers, employees and partners "exploring," or do they have more specific objectives in mind? Take a broader look and let the results speak for themselves.
Jason Hekl is vice president of corporate marketing at InQuira.